PaperlayerView exact revision on arXiv

Source-linked AI summary

Self-Control in Cyberspace: Applying Dual Systems Theory to a Review of Digital Self-Control Tools

Ulrik Lyngs, Kai Lukoff, Petr Slovak, Reuben Binns, Adam Slack, Michael Inzlicht, Max Van Kleek, Nigel Shadbolt

arXiv:1902.00157v1cs.HC

TL;DR

Understanding of the design mechanisms that support digital self-control remains limited, with no systematic review mapping the features of existing tools. The paper reviews 367 digital self-control apps and browser extensions through an adapted dual systems model, and reports that none of 17 related papers relied on dual systems models of self-regulation.

  • Problem

    No systematic review had mapped the design features of the many existing digital self-control tools, leaving understanding of how best to support self-control limited.

    Source: INTRODUCTION · p. 2

  • Method

    The authors review 367 digital self-control apps and browser extensions, using an adapted integrative dual systems model to organise and evaluate their features.

    Source: INTRODUCTION · p. 2, digital self-control tools · p. 3

  • Results

    None of the 17 reviewed papers on digital self-control interventions relied on dual systems models of self-regulation.

    Source: Theory Applied in Related Research · p. 3

  • Takeaways & Limitations

    Adding expected value of control to the model helps clarify how specific design features may work to scaffold successful self-control.

    Source: INTRODUCTION · p. 2

    System 2 control is limited by capacity and fluctuates with emotional state and fatigue.

    Source: Self-control limitations and the… · p. 5

Abstract

from arXiv · show

Many people struggle to control their use of digital devices. However, our understanding of the design mechanisms that support user self-control remains limited. In this paper, we make two contributions to HCI research in this space: first, we analyse 367 apps and browser extensions from the Google Play, Chrome Web, and Apple App stores to identify common core design features and intervention strategies afforded by current tools for digital self-control. Second, we adapt and apply an integrative dual systems model of self-regulation as a framework for organising and evaluating the design features found. Our analysis aims to help the design of better tools in two ways: (i) by identifying how, through a well-established model of self-regulation, current tools overlap and differ in how they support self-control; and (ii) by using the model to reveal underexplored cognitive mechanisms that could aid the design of new tools.

INTRODUCTION · 1 RELATED WORK · Evidence on Self-Control Challenges

The paper reviews digital self-control tools through an adapted dual systems model, addressing the limited understanding of their design features and the widespread difficulty people have controlling device use.

INTRODUCTION

  • Always-available devices can disrupt focus through notifications and habitual check-ins, while attention-economy business models incentivise frequent, extensive use.The paper frames these pressures as a downside of devices that provide access to many tasks anywhere and anytime.Source: INTRODUCTION · p. 1
  • HCI research has called for intentional non-use, while studies and a large app-and-extension market have offered interventions such as gamification, social sharing, and use visualisations.Examples include rewards for reducing smartphone use and visualisations of laptop use.Source: INTRODUCTION · p. 2
  • No systematic review had mapped design features across the hundreds of available tools, and dual systems and value-based models of self-regulation had not yet been applied to this problem.Earlier interventions had mainly drawn on user interviews, designer intuition, or theories such as cognitive load, Social Cognitive, and nudge theory.Source: INTRODUCTION · p. 2
  • The review analyses 367 digital self-control apps and browser extensions and applies an integrative dual systems model incorporating expected value of control to organise and evaluate their design features.The authors argue this approach can clarify how design features scaffold successful self-control.Source: INTRODUCTION · p. 2
  • Self-regulation refers broadly to processes serving goal-directed behaviour, including habits, whereas self-control refers more narrowly to deliberate efforts to override conflicting impulses.This distinction is given in the paper's terminology note.Source: INTRODUCTION · p. 2

Evidence on Self-Control Challenges

  • 64% of smartphone users surveyed by Ko et al. felt they were overusing their devices, and 60% wanted to change their usage habits.Desired changes involved frequent short check-ins disrupting tasks and excessive long use, such as habitual checking before bed.Source: Evidence on Self-Control Challenges · p. 2
  • Although severe self-regulation breakdowns in some users have been called addiction, current concern also encompasses daily struggles experienced by most users.The paper describes the broader pattern as difficulties with self-control over device use, rather than only severe cases.Source: Evidence on Self-Control Challenges · p. 2
  • Strategies often failed when users relied on willpower, because momentary impulses overrode their intentions to limit device use.Surveys, public discussion, and responses by technology companies suggest these day-to-day self-regulation struggles are common.Source: Evidence on Self-Control Challenges · p. 2

Theory Applied in Related Research · Digital Behaviour Change Interventions. · 2 AN INTEGRATIVE DUAL SYSTEMS APPROACH TO DIGITAL SELF-CONTROL

Prior HCI and digital behaviour-change research rarely uses dual systems theory, while reviews identify it as a potential framework for interventions targeting habitual behaviour. The paper therefore draws on dual systems theory to systematise and classify digital self-control tools.

Theory Applied in Related Research

  • The 17 HCI papers were selected from a literature collection of 500+ papers gathered over two years, but the collection was not a formal systematic review.The papers either built or evaluated digital self-control interventions; the analysis examined theories used to guide development or evaluate effects.Source: Theory Applied in Related Research · pp. 2–3
  • Seven of 17 HCI papers specified no self-regulation theory, the other 10 used varied models, and none relied on dual systems models [cf. 45, 145].The specified theories came from psychology, cognitive neuroscience, behavioural economics, behaviour change, and addiction research.Source: Theory Applied in Related Research · p. 3
  • HCI research on digital behaviour-change interventions (DBCIs) includes health applications addressing exercise, smoking cessation, diet, stress, and chronic conditions.The passage situates digital self-control tools as a subset of DBCIs.Source: Digital Behaviour Change Interventions · p. 3

Digital Behaviour Change Interventions.

  • In an 85-study DBCI review, the Transtheoretical Model appeared in 13/85 studies, Goal Setting Theory in 5/85, Social Conformity Theory in 3/85, and 51/85 specified no theory.None specified dual systems theory [cf. 145], and most studies that named theories did not explain how theoretical constructs informed intervention design or evaluation.Source: Digital Behaviour Change Interventions · p. 3
  • A recent DBCI review found that 60% (51/85) of studies did not specify a theoretical basis [cf. 160, 196], and none specified dual systems theory [cf. 145].Source: Digital Behaviour Change Interventions · p. 3
  • A DBCI review noted that most applied theories assume deliberative rationality, while arguing that dual systems theory could guide research on long-term change through breaking and forming habits [145, 178, 200; cf. 194].Source: Digital Behaviour Change Interventions · p. 3

2 AN INTEGRATIVE DUAL SYSTEMS APPROACH TO DIGITAL SELF-CONTROL

  • The paper draws on dual systems theory and argues that it can be used to systematise and classify digital self-control tools.Source: §2 · p. 3

digital self-control tools. · System 1 and System 2 · Action schema competition

The paper extends dual-systems theory with expected value of control, then uses the model to organize a review of digital self-control tools and identify design gaps. It explains self-regulation through competition between automatic and goal-directed control and the action schemas they activate.

digital self-control tools.

  • The authors extend dual-system applications in DBCIs with expected value of control, then apply the model to tools from three app stores.They use it to organize common design features and identify gaps and opportunities; expected value of control helps explain why self-control success fluctuates over time and with emotional state. Figure 1 depicts it as mediating System 2 strength.Source: digital self-control tools · p. 3, Figure 1 · p. 5

System 1 and System 2

  • Dual-systems theories distinguish swift, parallel, non-conscious System 1 processes from slower, capacity-limited, conscious System 2 processes.Source: System 1 and System 2 · p. 3
  • System 1 is driven by environmental inputs, internal states, and pathways linking situations to habits or instincts, enabling automatic behavior such as checking a smartphone for notifications [125, 134; cf. 28, 141, 189].Source: System 1 and System 2 · p. 3
  • System 2 is guided by goals, intentions, and rules in conscious working memory, supporting planning, decision-making, and resistance to habits or temptations.It also guides responses by coordinating sensory input, memory retrieval, emotional processing, and behavioral output.Source: System 1 and System 2 · pp. 3–4

Action schema competition

  • Action schemas are hierarchical control units for ordered action sequences that achieve goals, ranging from simple motor actions to complex tasks such as preparing tea.Source: Action schema competition · p. 4
  • Schemas compete for behavioral control according to activation strength; sensory input, higher-level schemas, and top-down System 2 control can activate them.Source: Action schema competition · p. 4

Self-regulation and self-control · Attentional filtering · Self-control limitations and the Expected Value of Control

The sections distinguish habitual self-regulation from deliberate self-control and explain how attention, limited capacity, fluctuating control, and expected value shape whether valued goals guide behavior.

Self-regulation and self-control

  • Self-regulation includes automatic goal-directed habits, whereas self-control is conscious, deliberate System 2 control when immediate impulses conflict with enduring goals.Source: Self-regulation and self-control · p. 5

Attentional filtering

  • Goals and intentions must enter working memory, where attentional filters favor high-activation signals; bottom-up cues such as notifications can win access automatically.Source: Attentional filtering · p. 5
  • System 2 can deliberately direct attention toward internal or external information, increasing its signal strength and likelihood of entering working memory.Source: Attentional filtering · p. 5

Self-control limitations and the Expected Value of Control

  • People may fail to follow valued goals despite recognizing the mismatch because System 2 control is capacity-limited and fluctuates with emotion and fatigue.Source: Self-control limitations and the… · p. 5
  • Limited working memory can leave goals unavailable at the moment of action, as feeds or inboxes can crowd out the purpose for opening an app.Source: Self-control limitations and the… · p. 5
  • System 2 control suffers from continuous exertion and varies with mood; negative mood predicts relapse, and Facebook users regulate platform time worse when feeling bad [25, 49, 73, 79, 186, 77, 127, 186, 156; cf. 130].Source: Self-control limitations and the… · pp. 5–6
  • Expected value of control (EVC) reflects a cost-benefit analysis and rises with perceived reward and confidence that self-control can achieve an outcome [28, 167, 3, 28, 142, 143, 192; 23; 180].Phone stack raises the stakes by adding financial and reputational costs to checking a phone.Source: Self-control limitations and the… · p. 6
  • EVC falls as the delay before the outcome of self-control grows, so a reward deferred for years is less motivating.Source: Self-control limitations and the… · p. 6

A practical example

A student repeatedly checks Facebook instead of beginning an essay, despite intending to work first. The model explains how context, goals, and expected value of control can shape this behavior.

A practical example

  • The student repeatedly scrolls Facebook when opening his laptop to work, despite intending to start the essay first and limit Facebook to breaks.He returns to the essay feeling regret after spending an inordinate amount of time on the news feed.Source: A practical example · p. 6
  • The model frames the situation through contextual cues, automatic System 1 behavior control, System 2 goals and self-monitoring, and expected value of control.Source: A practical example · p. 6
  • A habitual check may go unopposed if the essay goal is absent from working memory or expected value of control is low.Expected value may be low because inhibition offers no reward, confidence in suppressing the urge is low, or essay rewards are delayed.Source: A practical example · p. 6
  • After Facebook opens, attention-grabbing news-feed content can crowd the essay goal out of capacity-limited working memory.Source: A practical example · p. 6

3 A REVIEW AND ANALYSIS OF CURRENT DIGITAL SELF-CONTROL TOOLS · Methods · Initial Keyword Search and Data Clean Up.

The review searched app stores and a browser-extension store for digital self-control tools, then screened candidates and mapped included tools’ design features to a dual systems model.

3 A REVIEW AND ANALYSIS OF CURRENT DIGITAL SELF-CONTROL TOOLS

  • The review coded digital self-control apps and browser extensions and mapped their design features to components of the dual systems model.It covered Google Play, Apple App Store, and Chrome Web Store tools described as helping users exercise self-control, avoid distraction, or manage digital-device-use addiction.Source: §3 · p. 6

Methods

  • Google Play and Apple App Store searches used pre-existing scripts and terms including “distraction,” “addiction,” “motivation,” and “self-control,” with smartphone-related variants.The listed queries included “smartphone distraction,” “smartphone addiction,” “smartphone motivation,” and “smartphone self-control.”Source: Initial Keyword Search and Data Clean Up · p. 6

Initial Keyword Search and Data Clean Up.

  • For Chrome Web Store searches, the researchers built a scraper and used the same terms, replacing “smartphone” with “laptop” and adding “internet” variants.Examples included “laptop distraction” and “internet distraction.”Source: Initial Keyword Search and Data Clean Up · p. 6
  • Between August 22 and 27, 2018, searches of the US and UK stores yielded 4,890 distinct apps and extensions after duplicate results were removed.The total comprised 1,571 Google Play apps, 2,341 App Store apps, and 978 Chrome Web Store extensions.Source: Initial Keyword Search and Data Clean Up · pp. 6–7
  • Manual screening included apps and extensions explicitly designed to help people self-regulate digital-device use.Titles and short descriptions were screened; when unavailable, reviewers used the opening paragraphs of full descriptions.Source: Identifying Potentially Relevant Apps… · p. 7
  • Screening excluded tools for general productivity, self-regulation outside digital-device use, or tools unavailable in English.The paper points to its online supplementary material for detailed exclusion criteria.Source: Identifying Potentially Relevant Apps… · p. 7

Identifying Potentially Relevant Apps and Extensions. · Identifying Apps and Extensions to Analyse.

The review narrowed 731 potentially relevant apps and extensions to 380 for analysis, excluding duplicate iOS versions when an app appeared in both stores. Researchers then coded store-page functionality using an iteratively developed codebook and author review.

Identifying Potentially Relevant Apps and Extensions.

Identifying Apps and Extensions to Analyse.

  • 380 apps and extensions remained for analysis: 96 from Google Play, 60 from the App Store, and 224 from the Chrome Web store.Source: Identifying Apps and Extensions to… · p. 7
  • Because iOS restricts developer access to operating-system permissions more than Android, often limiting iOS app versions, the review excluded them to examine design-space coverage rather than platform differences.Source: Identifying Apps and Extensions to… · p. 7
  • Functionality was coded from store-page descriptions, screenshots, and videos, following similar reviews [cf. 166, 177–179].Source: Feature coding · p. 7
  • The iteratively developed coding sheet anticipated four feature clusters—block/removal, self-tracking, goal advancement, and reward/punishment—drawing on prior classifications [cf. 20, 140] and earlier work.Source: Feature coding · p. 7
  • Three authors independently classified features in 10 apps and 10 extensions before discussing categories; two authors then reviewed 60 additional tools each, and a third reviewed those 120 plus the remainder before final codebook recoding.Source: Identifying Apps and Extensions to… · p. 7
  • A tool could have multiple core design clusters, defined as any cluster representing at least 25% of its functionality.Source: Identifying Apps and Extensions to… · p. 7

Feature coding. · Results · Feature prevalence.

The final dataset contained 367 tools. Blocking or removing distractions was the most prevalent feature cluster, alongside tools for reducing exposure, tracking use, advancing goals, rewarding or punishing behavior, and defining distractions.

Feature coding.

  • The final dataset contained 367 tools after 13 were excluded during coding.The exclusions comprised 3 duplicates, 7 tools that had become inaccessible, and 3 without sufficiently described functionality for coding.Source: Feature coding · p. 7

Results

  • Blocking or removing distractions appeared in 74% of tools; 44% (163) added obstacles to distraction, and 14% (50) added friction to removing a block.Obstacles included blocking access, limiting time or launches, or delaying access; removal friction included making sessions non-stoppable, requiring effort or a password, restricting uninstallation, or delaying overrides and settings changes.Source: Feature prevalence · p. 7, Figure 3 · p. 8

Feature prevalence.

  • Reducing exposure to distracting options rather than blocking them appeared in 38% of tools (139), predominantly as Chrome extensions that removed distracting site elements.Among these tools, 67 removed specific-site elements, 27 were general reader extensions, 24 modified new tabs, and 22 supported minimal writing; Facebook, YouTube, Twitter, and Gmail were most frequently targeted.Source: Feature prevalence · p. 8
  • Self-tracking appeared in 38% of tools (139), including recording user history in 105, visualising captured data in 73, and displaying a timer or countdown in 72.In 42 tools, tracking focused on time spent not using digital devices.Source: Feature prevalence · p. 8
  • Goal advancement appeared in 35% of tools (130), through concrete-goal or task reminders in 58 tools, general-goal or value reminders in 52, and explicit goal setting in 58.Fifteen tools let users compare behavior against set goals; explicit goals concerned device-use time in 39 tools or focus tasks in 19.Source: Feature prevalence · p. 8
  • Reward or punishment features appeared in 22% of tools (80), most often points or streaks (41), leaderboards or social sharing (27), and achievements (21).Twenty tools represented points as a lifeform, while 10 offered real-world rewards or punishments, including financial loss, redeemable points, or self-administered electrical shocks.Source: Feature prevalence · p. 8
  • User control over what counted as distraction appeared in 35% of tools (129); among blocking tools, 101 used blacklists and 22 used whitelists.Blacklists blocked specified apps or sites while allowing others; whitelists allowed only specified apps or sites.Source: Feature prevalence · p. 8
  • Sixty-five percent of tools centered on one feature cluster, while 32% (117 tools) combined two; only Flipd and HabitLab combined all four clusters.Common pairings were blocking/removal with goal advancement (40 tools), self-tracking with reward/punishment (30), and blocking/removal with self-tracking (24).Source: Feature combinations · p. 8

Feature combinations. · Store comparison. · Mapping identified tool features to theory.

Digital self-control tools mainly prevent unwanted habits and responses, while their available strategies vary across stores; the paper maps these features to cognitive components of an integrative dual-systems model.

Feature combinations.

  • Browser extensions commonly minimise interface features, while mobile tools rely on blocking or restricting access because developers have less control over other apps’ displays.The authors attribute the difference to browser developers’ ability to modify HTML pages with CSS and JavaScript.Source: Figure 4 · p. 9, Store comparison · p. 9
  • HabitLab cycles through intervention types to learn which best helps users align internet use with their stated goals [cf. 97].Source: Feature combinations · p. 9
  • Scheduled app blocking is constrained on iOS because developers lack the required permissions, and the authors suggest more permissions could enable more granular controls.They note that some developers of popular anti-distraction tools have petitioned Apple for permissions.Source: Store comparison · p. 9

Store comparison.

Mapping identified tool features to theory.

  • Blocking habitual targets or their interface triggers can prevent habits from activating, while limits and redirection can scaffold new habits; goals, reminders, and usage tracking support conscious self-monitoring.Usage can be timed, recorded, visualised, and compared with goals.Source: Mapping identified tool features to… · p. 9
  • Rewards, punishments, value reminders, and motivational quotes target control’s reward component; timers and quotes can affect delay or expectancy, while blocking or removal makes unwanted responses unavailable.Source: Mapping identified tool features to… · pp. 9–10
  • 73% of tools target unwanted-habit prevention, compared with 18% that scaffold non-conscious habits and 23% that influence the delay component of control.The authors report that prevention makes unwanted responses unavailable, stopping them from winning action-schema competition.Source: Figure 6 · p. 10, Mapping identified tool features to… · p. 10

4 DISCUSSION · Research opportunities prompted by widely used or theoretically interesting design features

The review finds that digital self-control tools most often block distractions or remove interface features, while habit scaffolding and parts of expected-value-of-control are least targeted. The discussion uses widely adopted or theoretically interesting features to identify research opportunities, neglected design areas, and ethical questions.

4 DISCUSSION

  • 65% of the 367 reviewed tools focused on one design cluster, with blocking/removal most prevalent, followed by self-tracking, goal advancement, and reward/punishment.Another 32% focused on two clusters; feature frequencies differed across the Chrome Web Store, Play Store, and Apple App Store, likely reflecting developer-permission differences. Habit scaffolding was the least commonly targeted cognitive component, followed by delay and expectancy in expected value of control.Source: §4 · p. 10
  • The discussion highlights underexplored current features, gaps identified through the dual systems model, and the model’s potential to guide research and intervention design.It also outlines limitations and future work.Source: §4 · p. 10

Research opportunities prompted by widely used or theoretically interesting design features

  • Digital self-control tools offer hundreds of natural experiments, but many underlying design mechanisms still lack evaluation of efficacy and transferability.The authors present these approaches as promising starting points for research studies.Source: Research opportunities prompted by… · p. 10
  • Forest links device use to a virtual tree’s well-being, using abstention rather than action to influence the reward component of expected value of control.The paper describes Forest as the most popular example, with over 5 million Android users, and notes that numerous variations and clones exist.Source: Research opportunities prompted by… · p. 10
  • Timewarp redirects users from distracting websites to goal-aligned sites, resembling automated implementation intentions that may scaffold transfer from conscious System 2 goals to automatic System 1 habits.Implementation intentions are if-when rules linking a context to a desired response; digital behaviour change researchers have highlighted them as promising.Source: Research opportunities prompted by… · p. 10
  • Some tools add friction to overriding restrictions, as Focusly does with laborious keystrokes, raising design and ethical questions about holding users to past preferences [cf. 31, 112].Many tools both restrict access to distractions and make those restrictions difficult to override.Source: Research opportunities prompted by… · p. 10

Gaps identified by the dual systems model · Using the model directly to guide intervention research and design

The dual systems model highlights underexplored mechanisms in digital self-control and offers an adaptable framework for organising research and guiding psychologically informed design.

Gaps identified by the dual systems model

  • Habit scaffolding is the least-targeted cognitive component; implementation intentions and automation are candidate strategies, while four tools link blocking to location, including AppBlock and Bashful.The authors suggest location-linked blocking could automatically trigger a target behaviour in a desired context.Source: Gaps identified by the dual systems… · p. 11
  • Only 4% of tools targeted delay beyond timers, and just two of 367 directly used delays: Space lengthens iOS app launches and Pipe Clogger website launches.The authors note strong, reliable effects of delay sensitivity and sensitivity to online delays, and propose delay-based interventions as potentially effective.Source: Gaps identified by the dual systems… · p. 11
  • Expectancy was infrequently targeted, mainly through timers limiting attempted self-control; Wormhole Escaper lets users administer encouragement after suppressing an urge.Given self-efficacy’s role in self-regulation, the authors suggest such encouragement could boost confidence in the ability to exert self-control.Source: Gaps identified by the dual systems… · p. 11

Using the model directly to guide intervention research and design

  • Researchers can use the model to organise interventions by cognitive component and guide future studies, linking HCI design to wider psychological research on self-regulation.The model’s abstract nature supports use at different analytical levels to inspire research and specific design.Source: Using the model directly to guide… · p. 11
  • Designers can align choices with self-regulatory mechanisms and expand model components for theoretical detail; reward can be elaborated by stimulus type, timing, gains, and losses.The authors present the model as a starting point for design consideration, with components readily expandable when more theoretical detail is useful.Source: Using the model directly to guide… · p. 11
  • TimeAware used gain–loss sensitivity to guide its design and better supported productivity when visualising distracting rather than productive time; Adams et al. trialed perceptual biases in food-choice and voice-pitch interventions.The authors offer these HCI examples as designs informed by psychological theory.Source: Using the model directly to guide… · p. 11

Limitations and future work · 5 CONCLUSION

The paper analyzes 367 digital self-control apps and browser extensions and applies dual systems theory to evaluate their design. It also identifies limitations in the review and frames digital-device use as both a self-regulation challenge and a low-cost test bed for interventions.

Limitations and future work

  • The review focuses on tool functionality, leaving install numbers and user-review content for future work because of space limits and unavailable Apple App Store user counts.The authors note that this approach resembles related reviews.Source: Limitations and future work · p. 11
  • The dual systems model’s high-level formulation leaves its cognitive design space underspecified, although its components are grounded in basic self-regulation research and can be elaborated through lower-level theories.How directly interventions should be anchored in causal theories remains debated.Source: Limitations and future work · pp. 11–12
  • Digital devices pose an unprecedented self-regulation challenge through the many behavioural options and engaging activities they make instantly available.Their broad functionality also permits more precise, flexible, and lower-cost intervention tests than changing the physical environment, while relevant activity is readily measurable.Source: Limitations and future work · p. 12
  • Digital self-control research can serve as a test bed for interventions because many factors in device environments can be changed at minimal cost.Device-use context detection is more manageable than in digital behaviour-change research because much relevant activity can be measured.Source: Limitations and future work · p. 12

5 CONCLUSION

  • The paper contributes a comprehensive functionality analysis of digital self-control apps and browser extensions and applies a well-established self-regulation model to explain their design features.The review covers tools from the Google Play, Chrome Web, and Apple App stores.Source: §5 · p. 12
  • The authors hope the review’s 367 apps and extensions, viewed as natural experiments, and its dual systems model will help users build beneficial technology habits and resist predatorial nudging.The examples of predatorial nudging include clickbait advertisers and data harvesters.Source: §5 · p. 12

Related papers

Loading 1902.00157v1…